def perform(self):
# Search for good alphas
alphas = [10**-x for x in np.arange(-1, 9.01, 1.0)]
hiddens = [(50, ), (50, 100, 50)]
learning_rates = sorted([0.001, 0.005])
params = {
'MLP__activation': ['relu'],
'MLP__alpha': alphas,
'MLP__learning_rate_init': learning_rates,
'MLP__hidden_layer_sizes': hiddens
}
learner = MLPClassifier(max_iter=1000,
early_stopping=True,
random_state=self._details.seed,
shuffle=True,
verbose=self._verbose)
pipe = Pipeline([('MLP', learner)])
trainer.perform_experiment(ds=self._details.ds,
ds_name=self._details.ds_name,
clf_name='ANNClassifier',
params=params,
pipe=pipe,
seed=self._details.seed)
def perform(self):
params = {'DTClassifier__criterion': ['gini'],
'DTClassifier__max_depth': [20, 40, 60]} # , 'DT__max_leaf_nodes': max_leaf_nodes}
learner = DecisionTreeClassifier(random_state=self._details.seed)
pipe = Pipeline([
('DTClassifier', learner)])
trainer.perform_experiment(ds=self._details.ds, ds_name=self._details.ds_name,
clf_name='DTClassifier', params=params,
pipe=pipe,
seed=self._details.seed)
def perform(self):
params = {'LRClassifier__penalty': ['l2']}
learner = LogisticRegression(random_state=self._details.seed)
pipe = Pipeline([('LRClassifier', learner)])
trainer.perform_experiment(ds=self._details.ds,
ds_name=self._details.ds_name,
clf_name='LRClassifier',
params=params,
pipe=pipe,
seed=self._details.seed)
def perform(self):
random_forest = ensemble.RandomForestClassifier(
random_state=self._details.seed)
params = {"RF__max_depth": [3, 5, 7, 10, None],
"RF__n_estimators": [3, 5, 10, 25, 50, 150],
"RF__max_features": [4, 7, 15, 20]}
pipe = Pipeline([
('RF', random_forest)])
trainer.perform_experiment(ds=self._details.ds, ds_name=self._details.ds_name,
clf_name='RF', params=params, pipe=pipe,
seed=self._details.seed)
def perform(self):
booster = xgb.XGBClassifier(booster='gbtree')
params = {
'Xgb__n_estimators': [10, 20, 50, 150],
'Xgb__learning_rate': [0.02, 0.04],
'Xgb__max_depth': [20, 40]
}
pipe = Pipeline([('Xgb', booster)])
trainer.perform_experiment(ds=self._details.ds,
ds_name=self._details.ds_name,
clf_name='Xgb',
params=params,
pipe=pipe,
seed=self._details.seed)
def perform(self):
booster = GradientBoostingClassifier(random_state=self._details.seed)
params = {
'Boost__n_estimators': [10, 20, 50, 150],
'Boost__learning_rate': [0.02, 0.04],
'Boost__max_depth': [20, 40]
}
pipe = Pipeline([('Boost', booster)])
trainer.perform_experiment(ds=self._details.ds,
ds_name=self._details.ds_name,
clf_name='Boost',
params=params,
pipe=pipe,
seed=self._details.seed)
def perform(self):
neighbors = [10, 20, 30]
learner = KNeighborsClassifier()
pca = PCA(random_state=self._details.seed)
params = {
'pca__n_components': [15, 20],
'KNN__metric': ['manhattan', 'euclidean', 'chebyshev'],
'KNN__n_neighbors': neighbors,
'KNN__weights': ['uniform']
}
pipe = Pipeline([('pca', pca), ('KNN', learner)])
trainer.perform_experiment(ds=self._details.ds,
ds_name=self._details.ds_name,
clf_name='KNNClassifier',
params=params,
pipe=pipe,
seed=self._details.seed)